Network resource management

A procedure performed in one or more networks of communication packet comprising user devices and network elements, the method comprising the steps of: sending a management components resources of at least some of the devices automatically information upgrade resources to a server resource availability, and use the server availability of resources such update information resources to generate a predictive model for characteristics of network resources for communication sessions, and make available model data prediction and update information resources as output data server, where the server resource availability creates a prediction model of an access network to predict characteristics that affect a session between a network element or device in said access network and another network element or device; and wherein the server resource availability receives an update information including feedback data with respect to a renegotiation of session management components session quality of the devices and network elements and updates the model prediction consequence; wherein the prediction model includes a model of self-learning to learn the behavior of a network access over time; and wherein the prediction model includes a linear model that keeps track of current use; and wherein the output data server including the load factor and network congestion predicted; and wherein the output data server includes historical information associated with dates or times of day or location; wherein, when the prediction model is generated, the server calculates the load factor of the network using load factors individual network element or device as samples; and wherein the server models an adding based on an analysis of nearest neighbor intervals using IP address and / or geographical location and / or type of network access devices; and wherein the server calculates the load factor using filters, in which all the load factors of the individual devices form an input vector for a filter and the output is the load factor added to the network as a all; and wherein the server stores the load factors in an historical database and correlate historical data based on the time of day, day of the week, and the week of the year with the factor current load measured by filters, and the prediction model provides a short-term prediction based on a historical trend to indicate whether the quality is expected to deteriorate or improve; and wherein the server before modeling prediction, determines the extent to which the devices have a strong correlation with respect to the behavior of QoS each other, and applies a threshold to the correlation required for devices that are in the same network.